fix behavior while shrinking

Adjust mapsize as we're unmapping pages.  Do not lie to glibc about
shrinking by less than a page.  It's unnecessary because we are not
giving back any memory to the kernel, but also it forces us to zero
out this memory because morecore() assumes by default that "new"
memory is already zero'd.

Signed-off-by: Guillaume Morin <[email protected]>
Signed-off-by: Eric B Munson <[email protected]>

morecore.c

@@ -178,20 +178,37 @@ static void *hugetlbfs_morecore(ptrdiff_t increment)
 		if (ret) {
 			WARNING("Unmapping failed while shrinking heap: "
 				"%s\n", strerror(errno));
-		} else if (!__hugetlb_opts.map_hugetlb && !using_default_pagesize){
-
-			/*
-			 * Now shrink the hugetlbfs file.
-			 */
+		} else {
 			mapsize += delta;
-			ret = ftruncate(heap_fd, mapsize);
-			if (ret) {
-				WARNING("Could not truncate hugetlbfs file to "
-					"shrink heap: %s\n", strerror(errno));
+			/*
+			* the glibc assumes by default that newly allocated
+			* memory by morecore() will be zeroed.  It would be
+			* wasteful to do it for allocation so we only shrink
+			* the top by the size of a page.
+			*/
+			increment = heapbase - heaptop + mapsize;
+
+			if (!__hugetlb_opts.map_hugetlb && !using_default_pagesize){
+
+				/*
+				* Now shrink the hugetlbfs file.
+				*/
+				ret = ftruncate(heap_fd, mapsize);
+				if (ret) {
+					WARNING("Could not truncate hugetlbfs file to "
+						"shrink heap: %s\n", strerror(errno));
+				}
 			}
 		}
 
 	}
+	else if (increment < 0) {
+		/* Don't shrink by less than a page to avoid having to zero
+		 * the memory.  There is no point in lying to glibc since
+		 * we're not freeing any memory.
+		 */
+		increment = 0;
+	}
 
 	/* heap is continuous */
 	p = heaptop;
@@ -355,7 +372,7 @@ void hugetlbfs_setup_morecore(void)
 	/* Set some allocator options more appropriate for hugepages */
 
 	if (__hugetlb_opts.shrink_ok)
-		mallopt(M_TRIM_THRESHOLD, hpage_size / 2);
+		mallopt(M_TRIM_THRESHOLD, hpage_size + hpage_size / 2);
 	else
 		mallopt(M_TRIM_THRESHOLD, -1);
 	mallopt(M_TOP_PAD, hpage_size / 2);